My 5 cents on the physical evidence against Lance Armstrong:
After LA's statement that he would not challenge USADA’s accusations, and the statements that followed from USADA and WADA, every journalist and blogger who write about this saga appear to assume that LA gave in not just because of the possible witness statements but because USADA had come up with incontrovertible physical evidence.
I think that is wrong, and so did Judge Sparks, who stated that from what he understood from USADA's "woefully inadequate charging letter" (his words in the handed down decision) USADA would mainly rely on witness statements.
Also, the wording in USADA's letter to LA about the blood samples is true legalese:
Lance Armstrong's doping is further evidenced by the data from blood collections obtained by the UCI from Lance Armstrong from 2009 and 2010. This data is fully consistent with blood manipulation including EPO use and/or blood transfusions. [My emphasis]
To start with, the USADA states that they have obtained data (but not blood samples) from the UCI. Blood samples would have allowed further testing with new methods, but they are not saying that. What they actually are saying is that the agency has looked at data already analyzed by the UCI, and has come to the conclusion that blood data is ”consistent with blood/and or EPO doping”. However, USADA does not state whether the data is consistent with other conditions.
This post is going to try to answer that question.
The UCI data received by USADA consists of 38 blood samples from LA during 2008 – 2012. The data sheet has been posted on several websites, eg here where the pdf-file can be downloaded.
I cannot guarantee its provenance (and as I will discuss below there are reasons why I hope this is not an original data file from UCI and/or USADA), but let us analyze it as it is presented.
First some explanation of the data:
The first column is the sample code.
The second column is red cell count (RBC); the number of red cells per micro liter blood. This is a measured variable.
The third column is the concentration of hemoglobin (Hb, the red, oxygen carrying, pigment in the blood) given as grams/deciliter. This variable is also measured.
The fourth column is the hematocrit; the share of the blood volume made up of cells (mostly red blood cells), expressed as percent of the total volume. This variable is either measured or calculated as the product of the number of red blood cells and the MCV (see below).
The fifth column is the mean red blood cell volume (MCV); the average volume of a red blood cell, expressed as fentoliters (liters times ten to the power of minus fifteen).
The MCV is either measured or calculated from the ratio of hematocrit/number of red blood cells.
The sixth column is the mean weight of hemoglobin in a red blood cell (MCH); calculated as the ratio of hemoglobin concentration/number of red blood cells, expressed in picograms (grams times ten to the power of -12).
The seventh column is the mean cell hemoglobin concentration (MCHC), calculated as the ratio of hemoglobin concentration/ hematocrit; expressed in g/deciliters.
The eight column contains data on the ratio of reticulocytes (immature red blood cells) to the number of red blood cells, expressed in percentages. Variable calculated from the measured number of reticulocytes/ and total red blood cells (each variable the number of cells times ten to the power of 6 per microliters).
The ninth column is the number of reticulocytes (*10^6)per microliter. This is a measured variable.
The tenth column is the OFF-SCORE. This is a calculated variable, and is very important because high numbers (larger than 126, or in some sports federations 133) is considered an indication of blood doping or EPO use. The OFF-SCORE is calculated from the following formula:
OFF SCORE = Hb-concentration – 60 * square root (Ret %)
The last column is the date the sample was taken.
To summarize, the data file consists of 4 measured variables, RBC, Hb, Reticulocyte number, and hematcorit or MVC, and 5 calculated ratios from those measured variables.
The normal values for these variables are (varies somewhat between various laboratories):
RBC: 4.0 – 5.4 *10^6/uL
Hb: 140 – 170 g/L (or 14 – 17 g/dL)
Hct: 38.8 – 48.1%
MCV: 80 – 100 fL
MCH: 28 – 35 pg
MCHC: 32 – 36 g/dL
Ret %: 0.2 – 2%
Ret #: 10 – 100 *10^9/L = 0.01 – 0.1 * 10^6/uL
Off-Score: average ca 85, shall be less than 126
If we compare these ranges with Armstrong’s data we find that with the exception of the RET# column there are only two LOW Hb-values, one LOW Hct-value, and two HIGH MCHC-values that are outside the normal range. We will come back to these values, but let’s first look at column nine, the number of reticulocytes.
The lower normal range is 0.01*10^6/uL. All of Armstrong’s values are below the lower limit! If we take this at face value, then the whole issue is moot. LA may suffer from anaplastic anemia – maybe caused by the chemotherapy used to treat his testicular cancer. He would probably need cortisone treatment, blood transfusions and/or EPO to stay alive.
Do I believe this? NO!
Truly, LA would be a medical phenomenon if this were true. However, before we even begin to consider such outlandish theories, we should check the other variables dependent on the reticulocyte measurements. Let’s look at the first RET% data. It is calculated by dividing the RET# (0.00436*10^6) by the number of RBC (4.4*10^6) which is equal to 0.00099, which is 0.099% (and not 0.99% which is the corresponding number in column eight).
So, let Occham’s razor rule, and assume that someone made an error in the conversion of the reticulocyte data. The numbers in column nine should be multiplied by 10.
No big deal, is it? It is easy to make errors in ones Xls-sheets. I know! I have done many, many! But then you take a step back and consider the fact that these measurements are used to determine if athletes will be allowed to continue to compete, to continue with their livelihood. If data sheets are handled in this sloppy way, what about data collection? Worrisome, to say the least.
OK, so apart from some few LOW values, all Armstrong’s blood data is within the normal range, so why the suspicion?
According to news stories (and cycle forums) it seems that the pattern of LA’s values is what has caught suspicion. According to what appears to be a leak from the USADA, it is the specific pattern of LA’s Hcr and Hb during the end of the Giro 2009 and the start of the Tour the same year that is suspicious. To explain why this is so, let me explain what is thought to happen during blood doping and EPO.
When an athlete decides to blood dope he will have to withdraw some of his own blood. The volume is substituted with saline, causing the RBC, Hct, and Hb to decrease. Also, the Ret# increases (often to 2 – 3%), just as after any bleeding. When the blood is reinfused the Hb, RBC, and Hct will increase, but due to the sudden increase of RBC, the Ret% will decrease. A similar pattern is seen with EPO use. A few days after an EPO injection the Ret% increases (depending on dosage). However, at a later stage the homeostasis kicks in and the Ret% is reduced (sometimes to 0.2% or below, but most often to around 0.4%. (see SIAB)
The year 2009 the Giro took place 9 – 31 May, and the Tour 4 – 26 July. Let us have a look at LA’s test data for the relevant period:
Date/RBC/Hb/Hcr/Ret%/Off-score
7 May/4.55/148/43.5/1.33/78.3
18 May/4.19/136/40/0.72/85.1
31 May/4.1/130/38.2/0.89/73.4
It is obvious that LA’s blood values decreased during the Giro. Could this have been due to blood letting in preparation for the Tour? Possibly, but what about the reticulocyte percentage? Ought not the Ret% increase due to blood letting? (Yes, it ought to do that.)
Actually, one of the stated reasons for suspecting EPO doping (in small doses) during the major stage cycle races is that the blood values of the athletes do not fall as the races progress. A decrease in blood values is considered normal. So, what is considered a normal response in most athletes, is considered a suspicious finding for Armstrong.
Let us look at another racer, Wiggins, who posted some of his 2009 blood data on the web:
Date/Hb/Off-score
9 Apr/148/92
9 May/150/85
19 May/138/65
So, at the same stage of the competition (18 to 19th May) Wiggins had lost exactly the same Hb-concentration as LA, but Wiggin’s off-score decreased. Calculated Ret% for Wiggins the 9 May is 1.17, and for 19 May 1.48. Not big enough difference to cause alarm, but definitely in the ”right” direction for blood letting. Reaction? Crickets.
As will be shown below, during the subsequent Tour Lance’s blood values remained relatively constant. One should remember that Armstrong’s preparations for the Giro, including a fractured collar bone, were far from perfect, More about this later.
The LA data just prior to, during, and after the Tour (2009):
Date/RBC/Hb/Hcr/Ret%/Off-score
16 June/4.94/160/45.7/0.84/112
17 June/4.94/160/45/0.74/108.4
2 July/4.57/143/42.8/0.54/98.9
10 July/4.4/140/41.3/0.54/95.9
11 July/4.33/137/40.7/0.53/93.3
14 July/4.57/144/43.1/0.86/95.9
20 July/4.42/140/41.7/0.54/95.9
25 July/4.58/145/43/0.67/95.9
12 Aug/5.02/161/43.8/1.17/96.1
11 Dec/4.77/155/43.1/0.78/102
Here we have it. The data points that have been discussed are the 16th and 17th June. How could LA’s blood values increase so quickly? He must have cheated! Or?
Armstrong returned from Italy to attend the birth of his daughter(?). He then trained at altitude in Aspen (ca 8000 ft , 2300 – 2800 m altitude). Can altitude exposure cause such a fast recovery? No theorizing here. Let us have a look at some data.
In the early 90s the Swedish Defence Research Agency carried out a series of experiments on the use of hypoxic atmospheres in submarines. Subjects were exposed to low oxygen concentrations (15 – 13%) at normal surrounding pressures. The aim was to investigate whether low oxygen atmospheres could be used aboard submarines to reduce the fire hazard. The experiments took place in a hyperbaric saturation chamber complex where the subjects were staying for 10 to 14 days. (It should go without saying that in this setting it would be impossible for the subjects to dope themselves without the connivance of the research team.) During the time in the chamber the subjects’ blood values were analyzed. Here are some results:
Series I (Hypoxic level 15%)
Hours in hypoxia/RBC/Hb/Hct/Ret%/Off-score 0/5.2/155/45/1.98/70.6
54/5.5/165/49/2.32/73.6
90/5.5/169/50/2.29/78.2
Series II (Hypoxic level 14%)
Days in hypoxia/RBC/Hb/Hct/Ret%/Off-score 0/5.1/154/44/0.71/103.4
5/5.3/159/46/0.81/105
10/5.5/168/49/0.33/133.5
(NB: Figures are averages of 8 subjects/group. Ret% and Off-Scores are calculated from the averages given in tables. For anyone who has managed to read this far and are interested in the references to the reports and full data set, please, Freep-mail me.)
The effective inspired oxygen partial pressures at 2300 – 2800 m asl are 15.1 – 14.0 kPa. The corresponding partial pressures at sea level breathing a mix of 15% oxygen is 14.3 kPa and 14% oxygen 13.3 kPa, so performing hard physical exercise at the altitude of Aspen would compare quite well to the normobaric oxygen exposures.
The results show clearly that a period less than half a week in hypoxia is enough to induce a substantial increase in circulating red blood cells. For subjects exposed to 14% oxygen for 10 days there was a definite decrease in circulating reticulocytes and consequently an increase in calculated off-score.
The erythropietin levels (EPO) increased on average 2 – 3 times after one day in hypoxia. The peak EPO occurred after 2 days, then the EPO levels declined, but still remained some 50% above the control level.
The plasma iron (ferritin) levels declined at the end of the experiment and this is probably the reason why the reticulocyte number also decreased. The body’s iron stores cannot keep up with the rapid production of new blood cells.
It should not be necessary to point out the similarity between these results and Armstrong’s blood test post his altitude training sojourn.
Finally, we have the data for the blood tests taken during the Tour de France 2009. As noted previously the data are all well within the normal limits. A Danish researcher has pointed to, what he considers an anomaly, when Armstrong’s Hb value increased from 137 to 144 g/L from the 11 to the 14 July. However, avoiding discussion of any possible physiological variation, the fact is that the measurement error is about 1%. Add to that the blood sampling error, which is of the same magnitude, and you find that there is no statistical significant difference between the two values.
The discussion whether it is abnormal for the blood values to remain constant during week long bicycle races belongs to the scientific literature, and not to the anti-doping agencies. Some researchers have shown the Hb, and Hcr values decline during prolonged periods of physical exercise, but this is not a universal observation. Likewise, the mechanism for an exercise induced decline in Hb is not determined. Possible "culprits" are exercise induced destruction of red blood cells, reduction in iron stores, or fluid shifts in the body. Well trained individuals appear less affected than sedentary individuals. (Some references to statements by haematologists, and scientific papers: 1, 2, 3, 4, 5. NB: The first reference is in danish, and the second in norwegian. Use web-translators.)
BTW, once again Armstrong’s blood values compare well with Wiggins’s (Not my intention to pick on the latter, but his data is available):
Date/Hb/Off-score
2 July/150/93
14 July/140/85
20 July/152/95
25 July/139/83
The Hb value during the 2nd rest-day (20 July) is definitely more ”off” than LA’s, but again no reactions from the cycling world.
(No, I don’t think there is any evidence that Wigins was doped. Fluid shifts can create large and fast differences in blood values. For example, just by lying down the Hb-value will decrease by 8% in a normal individual. In healthy individuals it takes about 30 min for the body fluids to equilibrate after a change in body position.
And a link to a note by dr Michele Ferrari (I know, not the darling of the anti-doping agencies… but this is a MUST READ.). I haven’t seen these data published elsewhere, and I can understand that the WADA, USADA etc wouldn’t like to see them in print, but how would you like your livelihood depend on tests with this variability?)
Finally, I promised to come back to the last two data points that were out of range in LA’s data set.
The MCHC values for 12 August, and 11 December are suspiciously large. Again, this would indicate disease, rather than doping. MCHC is the ratio of Hb-concentration to hematocrit. Plotting the number of red blood cells vs hemoglobin for LA’s data set yields a straight line relationship, just as expected. Plotting the number of red blood cells vs hematocrit also yields a straight line relationship, except for two data points (12 August and 11 December). The Hcr values for these days are too low (giving rise to a large MCHC number). For the blood passport only Hb and reticulocyte numbers are used. Nevertheless, it should have been obvious to the testing agency that those Hcr values were off, and either a new measurement from the same blood sample could have been performed, or at least it should have been noted that the quality of those measurements were suspect.
Michele Ferrari quoted the French cyclist Jacques Anquetil saying:
"As a principle, I refuse to subject myself to such controls… I don’t want one of my victories to be put in doubt by the fantasy of the analyses”.
Unfortunately, judging from the way Lance Armstrong has been handled, I must admit the old Frenchman has a point.
Apart, from anything else, it looks like Armstrong has found himself in the middle of a turf war between the international doping agency and USADA. This article indirectly explains why it is so important for the USADA to ”nail” Lance Armstrong.
In conclusion, I cannot find anything in the published blood data that proves that Lance Armstrong used blood doping or EPO injections. The data points from the end of the Giro until the beginning of the Tour may be consistent with EPO use (but not blood doping), but are just as consistent with altitude training.
